Association between corticosteroids and in-hospital mortality after cardiac arrest: A multicenter study using the eICU Collaborative Research Database.

BACKGROUND

Clinical outcomes following cardiac arrest remain poor, and the therapeutic value of corticosteroids is controversial. We investigated the association between corticosteroid administration and in-hospital mortality across clinical phenotypes.

METHODS

Using the eICU Collaborative Research Database, we included 4,580 adults admitted to the ICU post-cardiac arrest. Exposure was defined as systemic corticosteroid administration during the ICU stay. We employed multivariable logistic regression, propensity score matching, and multiple imputation to adjust for baseline confounders. A 24-hour landmark analysis addressed immortal time bias.

RESULTS

Corticosteroid administration was independently associated with lower in-hospital mortality (adjusted OR 0.67; 95% CI 0.56-0.81; P < 0.001), remaining robust across propensity-matched and landmark analyses. Significant effect heterogeneity emerged across phenotypes. Survival benefits were exclusively observed in comatose patients (Glasgow Coma Scale ≤ 8) and those with comorbid chronic obstructive pulmonary disease, whereas non-comatose patients derived no benefit. Notably, while corticosteroids were associated with higher rates of favorable functional neurological recovery, mortality benefits did not significantly differ across systemic inflammatory marker strata.

CONCLUSIONS

Corticosteroid exposure was associated with lower in-hospital mortality among post-cardiac arrest patients, with benefits predominantly observed in comatose phenotypes. These hypothesis-generating findings underscore the importance of phenotype-guided approaches in future prospective trials.